Abstract
DPYD-guided dosing has improved the safety of fluoropyrimidine-based chemotherapy in recent years. However, severe toxicity remains in ~ 23% of patients not carrying DPYD variant alleles treated with capecitabine. Therefore, we developed a predictive model based on patient-related and treatment-related factors aimed at estimating the risk of developing severe capecitabine-related toxicity. The nomogram was developed using data from two large clinical trials (NCT00838370 and NCT02324452). Patients with cancer carrying a DPYD variant allele (DPYD*2A, c.1236G>A, c.2846A>T, and c.1679T>G) were excluded. Univariable and multivariable logistic regression using predetermined predictors based on previous findings, including age, sex, body surface area, type of treatment regimen, and creatinine levels were used to develop the nomogram. The developed model was internally validated using bootstrap resampling and cross-validation. This model was not externally or clinically validated. A total of 2,147 DPYD wild-type patients with cancer treated with capecitabine-based chemotherapy regimens were included of which complete data of 1,745 patients were available and used for the development of the nomogram. Univariable and multivariable logistic regression showed that age, sex, and type of treatment regimen were strong predictors of severe capecitabine-related toxicity in DPYD wild-type patients. Internal validation demonstrated a concordance index of 0.68 which indicates a good discriminative ability for prediction of severe capecitabine-related toxicity. The developed nomogram includes readily available parameters and may be a helpful tool for clinicians to assess the risk of developing severe capecitabine-related toxicity in patients without known risk DPYD variant alleles treated with capecitabine-based anticancer regimens.